We have all had that moment when we look at some cute pet whining at us and we wonder what exactly is that dog or cat trying to say. Now, you can talk to animals using AI & technical advancements.
Wouldn’t it be wonderful if you could really learn their language just like you learn french or mandarin? Oh sure, but is that even possible?
What if I told you that we might already be achieving this feat? Yeah, no kidding! AI may help us realize this extraordinary act.
How you may ask? We have to go through a step by step process to understand this.
AI helped us in decoding ancient languages
Do you know how AI works in our everyday tech usage? Every time any of us use a tool such as Gmail, an AI in the background gets better at understanding languages. It analyses sentence structures, understands words choices, and recognizes idioms.
We have reached a point where w can actually decode languages not even spoken by alive people.
Researchers from Google and MIT have successfully applied these abilities to ancient scripts.
How is it done?
This has been achieved through a step by step process. The first step is to map word-to-word relations for a specific language using a vast database.
The system searches the frequency of a particular word appearing next to every other word. This forms a pattern of appearance, a unique signature. This pattern defines the word in multidimensional parameter space.
According to the estimates of scientists, all languages have about 600 independent dimensions of relationships. In these dimensions, every word-to-word relation can be seen as a vector in this space.
This vector acts as a powerful limitation in the way a word can appear in any translation that the machine comes up with.
These vectors follow basic principles like king-man + woman = queen.
We can describe each sentence as a set of vectors which forms trajectory through the word space.
How it may help us in understanding animals
To understand how scientists may apply the same principles in understanding animal language, we have to go on assuming some things for now.
Just imagine that whale songs are communicating in a word like structure. Now if the relationships that whales have for their ideas are similar to ours, then?
Then we can map the important elements of whale song to dimensional spaces. This can help us comprehend whatever it is that whales are trying to say and help us communicate with them.
We don’t know yet what are the key elements of whale songs or elephant sounds, but this is beginning.
The advancement in AI tools recognizing these patterns will surely achieve the desired goal soon.
